llvm-mc.cpp revision 519466c32fe08bdf9afabfabd60a42f6c827ea6b 
1//===-- llvm-mc.cpp - Machine Code Hacking Driver -------------------------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This utility is a simple driver that allows command line hacking on machine
11// code.
12//
13//===----------------------------------------------------------------------===//
14
15#include "llvm/MC/MCParser/MCAsmLexer.h"
16#include "llvm/MC/MCContext.h"
17#include "llvm/MC/MCCodeEmitter.h"
18#include "llvm/MC/MCInstPrinter.h"
19#include "llvm/MC/MCSectionMachO.h"
20#include "llvm/MC/MCStreamer.h"
21#include "llvm/MC/MCParser/AsmParser.h"
22#include "llvm/Target/TargetAsmBackend.h"
23#include "llvm/Target/TargetAsmParser.h"
24#include "llvm/Target/TargetData.h"
25#include "llvm/Target/TargetRegistry.h"
26#include "llvm/Target/TargetMachine.h"  // FIXME.
27#include "llvm/Target/TargetSelect.h"
28#include "llvm/ADT/OwningPtr.h"
29#include "llvm/Support/CommandLine.h"
30#include "llvm/Support/FormattedStream.h"
31#include "llvm/Support/ManagedStatic.h"
32#include "llvm/Support/MemoryBuffer.h"
33#include "llvm/Support/PrettyStackTrace.h"
34#include "llvm/Support/SourceMgr.h"
35#include "llvm/Support/raw_ostream.h"
36#include "llvm/System/Host.h"
37#include "llvm/System/Signals.h"
38#include "Disassembler.h"
39using namespace llvm;
40
41static cl::opt<std::string>
42InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
43
44static cl::opt<std::string>
45OutputFilename("o", cl::desc("Output filename"),
46               cl::value_desc("filename"));
47
48static cl::opt<bool>
49ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
50
51static cl::opt<bool>
52ShowInst("show-inst", cl::desc("Show internal instruction representation"));
53
54static cl::opt<unsigned>
55OutputAsmVariant("output-asm-variant",
56                 cl::desc("Syntax variant to use for output printing"));
57
58static cl::opt<bool>
59RelaxAll("mc-relax-all", cl::desc("Relax all fixups"));
60
61enum OutputFileType {
62  OFT_Null,
63  OFT_AssemblyFile,
64  OFT_ObjectFile
65};
66static cl::opt<OutputFileType>
67FileType("filetype", cl::init(OFT_AssemblyFile),
68  cl::desc("Choose an output file type:"),
69  cl::values(
70       clEnumValN(OFT_AssemblyFile, "asm",
71                  "Emit an assembly ('.s') file"),
72       clEnumValN(OFT_Null, "null",
73                  "Don't emit anything (for timing purposes)"),
74       clEnumValN(OFT_ObjectFile, "obj",
75                  "Emit a native object ('.o') file"),
76       clEnumValEnd));
77
78static cl::opt<bool>
79Force("f", cl::desc("Enable binary output on terminals"));
80
81static cl::list<std::string>
82IncludeDirs("I", cl::desc("Directory of include files"),
83            cl::value_desc("directory"), cl::Prefix);
84
85static cl::opt<std::string>
86ArchName("arch", cl::desc("Target arch to assemble for, "
87                            "see -version for available targets"));
88
89static cl::opt<std::string>
90TripleName("triple", cl::desc("Target triple to assemble for, "
91                              "see -version for available targets"));
92
93static cl::opt<bool>
94NoInitialTextSection("n", cl::desc(
95                   "Don't assume assembly file starts in the text section"));
96
97enum ActionType {
98  AC_AsLex,
99  AC_Assemble,
100  AC_Disassemble
101};
102
103static cl::opt<ActionType>
104Action(cl::desc("Action to perform:"),
105       cl::init(AC_Assemble),
106       cl::values(clEnumValN(AC_AsLex, "as-lex",
107                             "Lex tokens from a .s file"),
108                  clEnumValN(AC_Assemble, "assemble",
109                             "Assemble a .s file (default)"),
110                  clEnumValN(AC_Disassemble, "disassemble",
111                             "Disassemble strings of hex bytes"),
112                  clEnumValEnd));
113
114static const Target *GetTarget(const char *ProgName) {
115  // Figure out the target triple.
116  if (TripleName.empty())
117    TripleName = sys::getHostTriple();
118  if (!ArchName.empty()) {
119    llvm::Triple TT(TripleName);
120    TT.setArchName(ArchName);
121    TripleName = TT.str();
122  }
123
124  // Get the target specific parser.
125  std::string Error;
126  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
127  if (TheTarget)
128    return TheTarget;
129
130  errs() << ProgName << ": error: unable to get target for '" << TripleName
131         << "', see --version and --triple.\n";
132  return 0;
133}
134
135static int AsLexInput(const char *ProgName) {
136  std::string ErrorMessage;
137  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename,
138                                                      &ErrorMessage);
139  if (Buffer == 0) {
140    errs() << ProgName << ": ";
141    if (ErrorMessage.size())
142      errs() << ErrorMessage << "\n";
143    else
144      errs() << "input file didn't read correctly.\n";
145    return 1;
146  }
147
148  SourceMgr SrcMgr;
149
150  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
151  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
152
153  // Record the location of the include directories so that the lexer can find
154  // it later.
155  SrcMgr.setIncludeDirs(IncludeDirs);
156
157  const Target *TheTarget = GetTarget(ProgName);
158  if (!TheTarget)
159    return 1;
160
161  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
162  assert(MAI && "Unable to create target asm info!");
163
164  AsmLexer Lexer(*MAI);
165
166  bool Error = false;
167
168  while (Lexer.Lex().isNot(AsmToken::Eof)) {
169    switch (Lexer.getKind()) {
170    default:
171      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
172      Error = true;
173      break;
174    case AsmToken::Error:
175      Error = true; // error already printed.
176      break;
177    case AsmToken::Identifier:
178      outs() << "identifier: " << Lexer.getTok().getString() << '\n';
179      break;
180    case AsmToken::String:
181      outs() << "string: " << Lexer.getTok().getString() << '\n';
182      break;
183    case AsmToken::Integer:
184      outs() << "int: " << Lexer.getTok().getString() << '\n';
185      break;
186
187    case AsmToken::Amp:            outs() << "Amp\n"; break;
188    case AsmToken::AmpAmp:         outs() << "AmpAmp\n"; break;
189    case AsmToken::Caret:          outs() << "Caret\n"; break;
190    case AsmToken::Colon:          outs() << "Colon\n"; break;
191    case AsmToken::Comma:          outs() << "Comma\n"; break;
192    case AsmToken::Dollar:         outs() << "Dollar\n"; break;
193    case AsmToken::EndOfStatement: outs() << "EndOfStatement\n"; break;
194    case AsmToken::Eof:            outs() << "Eof\n"; break;
195    case AsmToken::Equal:          outs() << "Equal\n"; break;
196    case AsmToken::EqualEqual:     outs() << "EqualEqual\n"; break;
197    case AsmToken::Exclaim:        outs() << "Exclaim\n"; break;
198    case AsmToken::ExclaimEqual:   outs() << "ExclaimEqual\n"; break;
199    case AsmToken::Greater:        outs() << "Greater\n"; break;
200    case AsmToken::GreaterEqual:   outs() << "GreaterEqual\n"; break;
201    case AsmToken::GreaterGreater: outs() << "GreaterGreater\n"; break;
202    case AsmToken::LParen:         outs() << "LParen\n"; break;
203    case AsmToken::Less:           outs() << "Less\n"; break;
204    case AsmToken::LessEqual:      outs() << "LessEqual\n"; break;
205    case AsmToken::LessGreater:    outs() << "LessGreater\n"; break;
206    case AsmToken::LessLess:       outs() << "LessLess\n"; break;
207    case AsmToken::Minus:          outs() << "Minus\n"; break;
208    case AsmToken::Percent:        outs() << "Percent\n"; break;
209    case AsmToken::Pipe:           outs() << "Pipe\n"; break;
210    case AsmToken::PipePipe:       outs() << "PipePipe\n"; break;
211    case AsmToken::Plus:           outs() << "Plus\n"; break;
212    case AsmToken::RParen:         outs() << "RParen\n"; break;
213    case AsmToken::Slash:          outs() << "Slash\n"; break;
214    case AsmToken::Star:           outs() << "Star\n"; break;
215    case AsmToken::Tilde:          outs() << "Tilde\n"; break;
216    }
217  }
218
219  return Error;
220}
221
222static formatted_raw_ostream *GetOutputStream() {
223  if (OutputFilename == "")
224    OutputFilename = "-";
225
226  // Make sure that the Out file gets unlinked from the disk if we get a
227  // SIGINT.
228  if (OutputFilename != "-")
229    sys::RemoveFileOnSignal(sys::Path(OutputFilename));
230
231  std::string Err;
232  raw_fd_ostream *Out = new raw_fd_ostream(OutputFilename.c_str(), Err,
233                                           raw_fd_ostream::F_Binary);
234  if (!Err.empty()) {
235    errs() << Err << '\n';
236    delete Out;
237    return 0;
238  }
239
240  return new formatted_raw_ostream(*Out, formatted_raw_ostream::DELETE_STREAM);
241}
242
243static int AssembleInput(const char *ProgName) {
244  const Target *TheTarget = GetTarget(ProgName);
245  if (!TheTarget)
246    return 1;
247
248  std::string Error;
249  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, &Error);
250  if (Buffer == 0) {
251    errs() << ProgName << ": ";
252    if (Error.size())
253      errs() << Error << "\n";
254    else
255      errs() << "input file didn't read correctly.\n";
256    return 1;
257  }
258
259  SourceMgr SrcMgr;
260
261  // Tell SrcMgr about this buffer, which is what the parser will pick up.
262  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
263
264  // Record the location of the include directories so that the lexer can find
265  // it later.
266  SrcMgr.setIncludeDirs(IncludeDirs);
267
268
269  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
270  assert(MAI && "Unable to create target asm info!");
271
272  MCContext Ctx(*MAI);
273  formatted_raw_ostream *Out = GetOutputStream();
274  if (!Out)
275    return 1;
276
277
278  // FIXME: We shouldn't need to do this (and link in codegen).
279  OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName, ""));
280
281  if (!TM) {
282    errs() << ProgName << ": error: could not create target for triple '"
283           << TripleName << "'.\n";
284    return 1;
285  }
286
287  OwningPtr<MCCodeEmitter> CE;
288  OwningPtr<MCStreamer> Str;
289  OwningPtr<TargetAsmBackend> TAB;
290
291  if (FileType == OFT_AssemblyFile) {
292    MCInstPrinter *IP =
293      TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI);
294    if (ShowEncoding)
295      CE.reset(TheTarget->createCodeEmitter(*TM, Ctx));
296    Str.reset(createAsmStreamer(Ctx, *Out,TM->getTargetData()->isLittleEndian(),
297                                /*asmverbose*/true, IP, CE.get(), ShowInst));
298  } else if (FileType == OFT_Null) {
299    Str.reset(createNullStreamer(Ctx));
300  } else {
301    assert(FileType == OFT_ObjectFile && "Invalid file type!");
302    CE.reset(TheTarget->createCodeEmitter(*TM, Ctx));
303    TAB.reset(TheTarget->createAsmBackend(TripleName));
304    Str.reset(createMachOStreamer(Ctx, *TAB, *Out, CE.get(), RelaxAll));
305  }
306
307  AsmParser Parser(SrcMgr, Ctx, *Str.get(), *MAI);
308  OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(Parser));
309  if (!TAP) {
310    errs() << ProgName
311           << ": error: this target does not support assembly parsing.\n";
312    return 1;
313  }
314
315  Parser.setTargetParser(*TAP.get());
316
317  int Res = Parser.Run(NoInitialTextSection);
318  if (Out != &fouts())
319    delete Out;
320
321  // Delete output on errors.
322  if (Res && OutputFilename != "-")
323    sys::Path(OutputFilename).eraseFromDisk();
324
325  return Res;
326}
327
328static int DisassembleInput(const char *ProgName) {
329  const Target *TheTarget = GetTarget(ProgName);
330  if (!TheTarget)
331    return 0;
332
333  std::string ErrorMessage;
334
335  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename,
336                                                      &ErrorMessage);
337
338  if (Buffer == 0) {
339    errs() << ProgName << ": ";
340    if (ErrorMessage.size())
341      errs() << ErrorMessage << "\n";
342    else
343      errs() << "input file didn't read correctly.\n";
344    return 1;
345  }
346
347  return Disassembler::disassemble(*TheTarget, TripleName, *Buffer);
348}
349
350
351int main(int argc, char **argv) {
352  // Print a stack trace if we signal out.
353  sys::PrintStackTraceOnErrorSignal();
354  PrettyStackTraceProgram X(argc, argv);
355  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
356
357  // Initialize targets and assembly printers/parsers.
358  llvm::InitializeAllTargetInfos();
359  // FIXME: We shouldn't need to initialize the Target(Machine)s.
360  llvm::InitializeAllTargets();
361  llvm::InitializeAllAsmPrinters();
362  llvm::InitializeAllAsmParsers();
363  llvm::InitializeAllDisassemblers();
364
365  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
366
367  switch (Action) {
368  default:
369  case AC_AsLex:
370    return AsLexInput(argv[0]);
371  case AC_Assemble:
372    return AssembleInput(argv[0]);
373  case AC_Disassemble:
374    return DisassembleInput(argv[0]);
375  }
376
377  return 0;
378}
379
380